Why does this regexp matches to almost everything?

Question

I cant really explain but check out the following:

name=$1
pat="\b[0-9a-zA-Z_]+\b"

if [[ $name =~ $pat ]]; then
  echo "$name is ok as user name"
else
  echo "$name is not ok as user name"
  exit 1
fi

Test run:

./script test_user+
test_user+ is ok as user name

The username with a + sign should not match that regexp.

btw [your script works fine](http://ideone.com/nzagNv) **without any change** — anubhava, Mar 04 '16 at 19:39
I am curious to the amount of downvotes as well. It is a good question. — jkdba, Mar 04 '16 at 19:40
@anubhava: It doesn't here in my testing. It says a username with a `+` is OK, which is not desired behavior. `bash 4.3.42(1)-release` Perhaps it depends on the version? — Dark Falcon, Mar 04 '16 at 19:41
It doesnt work on my system though :( I don't know why down votes, we should raise the down votes bar probably as on HackerNews. — Istvan, Mar 04 '16 at 19:41
@anubhava You are supplying the user name on standard input, which the script ignores, not as an argument. — chepner, Mar 05 '16 at 17:06
I posted an answer below to show that `\b` actually works on Ubuntu and other systems as word boundary — anubhava, Mar 05 '16 at 17:17

score 5 · Accepted Answer · edited May 23 '17 at 11:50

5

First of all:

\b is a PCRE extension; it isn't available in ERE, which the =~ operator in bash's [[ ]] syntax uses.

(From Bash regex match with word boundary)

Second, you don't want word boundaries (\b) if you wish to force the entire string to match. You want to match the start (^) and end ($):

pat="^[0-9a-zA-Z_]+\$"

edited May 23 '17 at 11:50

Community

1
1

answered Mar 04 '16 at 19:36

Dark Falcon

43,592
5
83
98

1

Thank you very much, indeed this is the problem. – Istvan Mar 04 '16 at 19:39

score 0 · Answer 2 · answered Mar 04 '16 at 19:45

0

if you dont want word bondry (guessed as you are trying username match) please use

^[0-9a-zA-Z_]+$

answered Mar 04 '16 at 19:45

Suneer

106
1
5

anubhava · Answer 3 · 2016-03-04T20:21:10.253

Contrary to the OP's experience and other answer it seems \b is supported on Ubuntu 14.04, bash 4.3.11 as word boundary. Here is a sample:

re='\bb[0-9]+\b'

[[ 'b123' =~ $re ]] && echo "matched" || echo "nope"
matched

[[ 'b123_' =~ $re ]] && echo "matched" || echo "nope"
nope

Even \< and \> also work fine as word boundaries:

re='\<b[0-9]+\>'

[[ 'b123' =~ $re ]] && echo "matched" || echo "nope"
matched

[[ 'b123_' =~ $re ]] && echo "matched" || echo "nope"
nope

However support of \b is specific to certain OS only. e.g. on OSX following works as word boundary:

[[ 'b123' =~ [[:\<:]]b[0-9]+[[:\>:]] ]] && echo "matched" || echo "nope"
matched

[[ 'b123_' =~ [[:\<:]]b[0-9]+[[:\>:]] ]] && echo "matched" || echo "nope"
nope

Why does this regexp matches to almost everything?

3 Answers3